The primary source documents a bullshit benchmark that measures detection rates over time across leading AI models, with the dedicated section at https://github.com/petergpt/bullshit-benchmark?tab=readme-ov-file#3-detection-rate-over-time presenting comparative data. (petergpt/bullshit-benchmark, 2024)

Repository results state a clear divergence of Anthropic’s models from other major models including ChatGPT and Gemini, positioning Claude with the lowest bullshit rate according to the benchmark. (petergpt/bullshit-benchmark, 2024)

The source provides graphs and metrics limited to detection rate trends without additional claims on adoption or trust factors. (petergpt/bullshit-benchmark, 2024)